Insights into the evolution, biogeography and natural history of the acorn ants, genus Temnothorax Mayr (hymenoptera: Formicidae)
BMC Evolutionary Biology volume 17, Article number: 250 (2017)
Temnothorax (Formicidae: Myrmicinae) is a diverse genus of ants found in a broad spectrum of ecosystems across the northern hemisphere. These diminutive ants have long served as models for social insect behavior, leading to discoveries about social learning and inspiring hypotheses about the process of speciation and the evolution of social parasitism. This genus is highly morphologically and behaviorally diverse, and this has caused a great deal of taxonomic confusion in recent years. Past efforts to estimate the phylogeny of this genus have been limited in taxonomic scope, leaving the broader evolutionary patterns in Temnothorax unclear. To establish the monophyly of Temnothorax, resolve the evolutionary relationships, reconstruct the historical biogeography and investigate trends in the evolution of key traits, I generated, assembled, and analyzed two molecular datasets: a traditional multi-locus Sanger sequencing dataset, and an ultra-conserved element (UCE) dataset. Using maximum likelihood, Bayesian, and summary-coalescent based approaches, I analyzed 22 data subsets consisting of 103 ingroup taxa and a maximum of 1.8 million base pairs in 2485 loci.
The results of this study suggest an origin of Temnothorax at the Eocene-Oligocene transition, concerted transitions to arboreal nesting habits in several clades during the Oligocene, coinciding with ancient global cooling, and several convergent origins of social parasitism in the Miocene and Pliocene. As with other Holarctic taxa, Temnothorax has a history of migration across Beringia during the Miocene.
Temnothorax is corroborated as a natural group, and the notion that many of the historical subgeneric and species group concepts are artificial is reinforced. The strict form of Emery’s Rule, in which a socially parasitic species is sister to its host species, is not well supported in Temnothorax.
Ants, the most diverse eusocial insects, are among the most successful arthropod groups worldwide, inhabiting and often dominating a broad spectrum of terrestrial ecosystems . The subfamily Myrmicinae is a success story within a success story: it contains nearly half of all described ant species, and six of the top ten most species-rich genera [2, 3]. Temnothorax, with approximately 390 named species, is one of these hyper-diverse myrmicine groups .
While the biology of most species of Temnothorax is poorly known, several taxa are frequently used as model organisms for social insect behavior [5, 6] and ecology [7, 8]. Studies on this genus have led investigators to discover new forms of social learning  and inspired hypotheses about speciation and the origins of parasitism . Temnothorax has a primarily Holarctic distribution (see Fig. 1), but notable exceptions include several species in Sub-Saharan Africa  and many species in Mesoamerica, including the islands of the Caribbean . Generally, Temnothorax ants are encountered in the mesic forests of the Northern hemisphere at mid-to-high elevations, where they are typically found under rocks, in leaf litter, or as arboreal foragers. However, these ants have adapted to a broad variety of habitats, from arid deserts to tropical rain forests, and are found at elevations from sea level up to 4000 m. The workers of most species in the Holarctic region are diminutive, timid, slow moving, and cryptically colored. Because of their inconspicuous nature, these ants are often overlooked despite their broad geographical distribution. The islands of the Greater Antilles harbor many Temnothorax endemics, however, that provide an exception to this habitus: the islands Cuba and Hispaniola, for example, have more than thirty described endemic species [12, 13], which are large, brightly colored, and capable of delivering powerful stings .
Temnothorax ants are believed to be trophic generalists, and have been observed scavenging for a variety of food items, including dead insects [14, 15], honeydew on leaves without tending aphids [16, 17], honeydew from tended aphids and membracids , the axillary nectaries of bracken ferns  and Baccharis (an Asteraceae genus) , seeds [14, 19], and elaiosomes . In competitions at baits, they appear to be opportunistic foragers, often absconding as more aggressive species arrive, and occasionally insinuating themselves into baits dominated by other species (Prebus, pers. obs.). Recruitment of workers to food and nest sites often appears to involve tandem running [21, 22].
Nests of Temnothorax are generally small, often with fewer than 200 workers , which may be distributed among several satellite nests [24,25,26]. With the worker caste of most species measuring less than four millimeters in length, these ants often take advantage of small cavities for their nesting sites, such as crevices in rocks, hollow nut shells, dead twigs, or directly in the soil. Temnothorax colonies tend to be monogynous, with only one queen per nest. This is hardly the rule, however: this genus displays a remarkable diversity of sociometry, with many species being either functionally monogynous (with several queens present, but only one actively producing eggs), or facultatively polygynous, both within and among species [27,28,29,30,31,32].
Many species appear to be arboreal specialists, nesting in dead hollow twigs on live trees, under bark, in galls, or under the roots of epiphytes.
Several taxa have intriguing life histories: it appears that Temnothorax and its sister group, the Leptothorax genus group (LGG, composed of Formicoxenus, Harpagoxenus, and Leptothorax) are unusually prone to developing a set of interrelated lifestyles collectively known as ‘social parasitism’. In general, social parasites are species that depend upon a free-living social host to complete their life cycle. In social insects, this often takes the form of exploiting the host for nesting space, brood care, and trophallaxis. This feature is found broadly in the aculeate Hymenoptera , but is especially prevalent in the ants. However, all recorded parasitic species are known only from the subfamilies of the formicoid clade [1, 34, 35]. Several different forms of social parasitism occur within the ants :
Xenobiosis, in which a parasitic species depends upon the host for nutrition and shelter, but cares for its own brood.
Temporary parasitism, in which a parasitic species depends upon the host only during its founding stage, eventually co-opting the nest site of the host.
Dulosis, or slave-making, in which a parasitic species produces workers, but is usually unable to forage or care for its own brood. Such species depend upon a host throughout their lives, co-opting the nest site as in temporary parasitism, but requiring periodic raids on other host nests for additional host workers, which perform the quotidian tasks within the nest.
Inquilinism, in which a parasitic species depends completely upon a host species for nutrition, shelter and brood care throughout its life. Typically, these species do not produce a worker caste, but divert all of their energy into producing sexuals.
All but temporary social parasitism have been recorded from Temnothorax and the LGG, with a recent study estimating that dulosis evolved at least six times within this clade .
Despite recent taxonomically focused molecular phylogenies [37,38,39] and extensive taxonomic works [12, 31], Temnothorax remains widely renowned as one of the most taxonomically difficult ant genera : its status as a phylogenetically natural group has been repeatedly revised  and called into question . This is a product of multiple factors: subtle morphological variation on a global scale among groups of species, striking departures from the morphological norm in others, morphological convergence with distantly related genera , weak diagnostic characters for the genus as a whole, and regional treatments of species in the absence of global understanding of the clade structure (see Additional file 1 for a more detailed account of recent molecular phylogenetic work and the taxonomic history of Temnothorax). The current absence of an infra-generic classification system for Temnothorax is not for a lack of effort, though: past taxonomists erected five subgenera to compartmentalize the species diversity of this genus, none of which are currently recognized due to a lack of clear diagnostic characters . In this study, I estimate the phylogeny of Temnothorax on a global scale to gain insight into the structure of relationships within the genus and, furthermore, to verify that the genus is monophyletic as currently diagnosed. To maximize within-genus lineage diversity, I sampled broadly from the species groups that have been formally designated in the literature. I simultaneously examine several key aspects of the historical biogeography and evolution in the genus, which I present below.
One of the most persistent biases in the history of biogeography are the ‘out-of-the-Palearctic’ and the ‘out-of-the-Holarctic’ hypotheses, which have their origins the theories of Linnaeus and Buffon, respectively. For the ants, much progress has been done to elucidate biogeographic origins of most of the large subfamilies and several genera [2, 34, 43,44,45,46,47,48,49,50,51,52,53], but these biases persist for some of the common lineages in the Northern Hemisphere. For example, Bernard  proposed that the species of the soil-nesting Palearctic Temnothorax rottenbergi group were the most ‘primitive’ taxa, implying that the that the Palearctic was the center of origin for the genus. This ‘out of the Palearctic’ hypothesis has remained unchallenged, despite recent works that have recognized extremely divergent Temnothorax occurring elsewhere, for example the Neotropical species of the former genus Macromischa (referred to as the salvini clade, the sallei subclade [allardycei + purpuratus + sallei + splendens groups] and the iris sublclade [iris + versicolor groups] in this study). See Additional file 1 for a discussion of the Macromischa syndrome. To address this issue, I reconstructed the biogeographic history of the ‘core formicoxenines’ (CF, composed of Gauromyrmex, LGG, Temnothorax, and Vombisidris), paying special attention to dispersal between the Palearctic and the Nearctic, as well as between the islands of the Greater Antilles and the New World mainland.
There are several dispersal routes that have been proposed for movements between the Palearctic and the Nearctic. In this study, I rule out a priori the north Atlantic De Geer route between Northwestern Palearctic and the Northeastern Nearctic, which operated from 71 to 63 million years old (Ma) , as the origin of the CF is estimated at 46.7 to 59.3 Ma . The Thulean route, which may have been important for early movements in the CF between the Western Palearctic and the Eastern Nearctic, opened further south than the De Geer route, and was exposed briefly 57 Ma and again 56 Ma . Another route for early dispersal is between the Eastern Palearctic and Western Nearctic along an early exposure of Beringia, which first emerged for a brief period 66 Ma, again 58.5 Ma, and then was exposed intermittently from 56 Ma onward, with an extended closure from the late Miocene to the Pliocene [55, 56]. The history of land connections to the Greater Antilles, as currently understood, is much simpler: these islands have been isolated from North America since their emergence in the early Cretaceous ca. 120–125 Ma, and were briefly connected to South America along the Greater Antilles-Aves Ridge during the late Eocene to the early Oligocene, ca. 33–35 Ma . This leaves overwater dispersal from mainland North or Central America as the most likely scenario. Given the morphological diversity of the Greater Antillean species, I make a first attempt to investigate the geographic origin, timing, and number of dispersals to these islands.
Another topic of particular interest is the placement of the morphologically divergent socially parasitic taxa, formerly known as the genera Chalepoxenus, Myrmoxenus, and Protomognathus (referred to as the muellerianus, corsicus, and americanus species groups, respectively, in this study). Carlo Emery observed that insect social parasites are nearly always closely related to their hosts . This generalization has come to be known as Emery’s Rule, and tends to be interpreted in two ways. In the ‘loose’ form, social parasites are congeneric with their host species, exploiting similarities between themselves and their host. Having evolved in allopatry, they are not necessarily close relatives of their hosts. In the ‘strict’ form, social parasites are the sister taxon to their hosts, and are in some cases thought to be the products of sympatric speciation [59,60,61,62]. Despite evidence that the parasitic former genera noted above are nested within Temnothorax, the degree to which they are related to their hosts (whether they adhere to the ‘loose’ or ‘strict’ form of Emery’s Rule), and whether they should retain status as valid genera remains a persistent topic of debate [63, 64].
The habitat under which ancestor of all ants inhabited has long been a topic of interest [65,66,67]. Long thought to have arisen from terrestrial habitats, the most widely accepted model of early evolution of the ants  suggests that they arose in the leaf litter of tropical angiosperm forests and subsequently spread to other strata. Recent work has largely supported this hypothesis [34, 47], but with the caveat that the most recent common ancestor of all crown ants was likely soil-dwelling . Yet this ‘out-of-the-ground’ hypothesis has often been extended to the arboreal species in some of the most species-rich genera of ants (e.g. Camponotus, Crematogaster, Polyrhachis) without much justification, as noted by Baroni Urbani . For example, Bernard  proposed that the Temnothorax rottenbergi group was plesiotypic within Temnothorax because the species nest directly in the soil, like many of the ant species in Northern and Western Europe (and in contrast to many of the Temnothorax species in this region). Until the present study, the implication that Temnothorax, which contains many arboreal species, is ancestrally ground-nesting and later transitioned to arboreal nesting habits has not been formally examined. A recent phylogeny of the Myrmicinae suggests that two arboreal lineages, Gauromyrmex and Vombisidris, are closely related to Temnothorax and its sister group, the LGG . In this study, I consider an alternative ‘out-of-the-trees’ hypothesis, in which Temnothorax arose from a terrestrially nesting ancestor, and subsequently spread to arboreal habitats.
Taxon sampling: Sanger dataset
An extended version of all methods can be found in Additional file 2. My taxon sampling strategy was designed to capture all the morphological, geographical, and life history variation within Temnothorax, with several exemplars chosen from each species group when possible. I selected 103 Temnothorax species, representing the recently synonymized satellite genera and 41 out of the 53 species groups currently in use, and 24 taxa of uncertain species-group affinity (see Additional file 3: Table S1 for a synopsis of the species groups). For outgroups, I used seven exemplars of the LGG, as well as representative species of the Indomalayan and Australasian genera Gauromyrmex and Vomibisidris, all of which have been shown to be close relatives of Temnothorax . To provide additional calibration points for divergence dating analyses and for rooting the tree, I added several more taxa which encompass the phylogenetic diversity of the subfamily Myrmicinae. Voucher specimens used for DNA extractions in this study were deposited in the University of California, Davis Bohart Museum of Entomology collection; unique specimen identifiers and collection data for these can be found in Additional file 4: Table S2 and at Antweb (http://antweb.org).
Sequence generation: Sanger dataset
DNA was extracted nondestructively or destructively from adult worker ants or pupae using a DNeasy Blood & Tissue Kit (Qiagen, Inc.) following the manufacturer’s protocols. Eight nuclear and two mitochondrial markers commonly used in ant systematics were chosen for amplification (for a complete list of primers used for amplification see Additional file 5: Table S3: 28S rDNA (28S), abdominal-A (abdA), arginine kinase (argK) including intron, elongation factor 1-alpha F2 (EF1aF2), long-wavelength rhodopsin (LW Rh) including intron, wingless (wg) including intron, rudimentary (CAD) including two introns, topoisomerase 1 (Top1), cytochrome c oxidase subunit I (COI) and cytochrome c oxidase subunit II (COII), including the spacer region between these two genes. Amplifications were performed using the PCR protocols described in Ward and Downie . Sequencing was performed using PCR primers and BigDye Terminator v3.1 Cycle Sequencing chemistry, and amplicons were analyzed on an ABI 3730 Capillary Electrophoresis Genetic Analyzer (Life Technologies) at the College of Biological Sciences DNA Sequencing Facility, University of California, Davis.
Processing and alignment: Sanger dataset
Sequence base calling and primer trimming was performed with Sequencher v 5.2.2 (Gene Codes Corporation) and exported for alignment as FASTA files. Because it has been suggested that arbitrary choice of alleles can bias phylogenetic results in some data sets , I retained ambiguous base calls (R or Y) for any potentially heterozygous sites. Sequences of the outgroup taxa Myrmica striolagaster, Solenopsis xyloni, Pogonomyrmex subdentatus, Tetheamyrma subspongia, Harpagoxenus sublaevis, Formicoxenus diversipilosus and Vombisidris bilongrudi were retrieved from Genbank. Identifiers for these and all new sequences generated by this study are listed in Additional file 6: Table S4.
Because intron sequence in the outgroup taxa diverged strongly from the ingroup, I maximized the number of informative sites for ingroup taxa by removing the introns from all outgroup taxa prior to alignment, retaining introns only in Temnothorax. Sequences were aligned with MAFFT v7.273 using the L-INS-I algorithm . Alignment confidence was assessed with the program ZORRO ; sites that fell below an arbitrary threshold of five were trimmed from the alignment using a custom Python script developed by Fernández et al. . Chi-squared test for nucleotide homogeneity were calculated with BaCoCa v 1.1.r . Because the third codon position of COI + COII failed the chi-squared test (P < 0.01); it was RY-coded for downstream analyses.
Phylogenetic inference: Sanger dataset
I used PartitionFinder 1.1.1  to select data partitions and simultaneously estimate best fitting substitution models for downstream maximum likelihood (ML) and Bayesian inference (BI) analyses. ML analyses were carried out with the programs RAxML v8.2.4  and IQ-TREE , using both best tree and bootstrap searches. BI was performed on the full dataset using MrBayes 3.2.6 , using the reversible jump Markov chain Monte Carlo algorithm. I ran MrBayes with two independent runs for 50 million generations, sampling every 5000 generations, and summarizing the run with a consensus tree after discarding a burnin of 25%. MCMC convergence was diagnosed by examining PRSF and average standard deviation of split frequencies between runs in the MrBayes ‘.stat’ output files. I also inspected trace plots and ESS with the program Tracer v1.6  for signs of adequate parameter mixing.
To investigate the possibility of gene tree-species tree conflict, I performed single gene BI analyses with MrBayes using the same settings as for the full concatenated analysis, but reducing the number of generations and sampling frequency to 10 million and 1000, respectively. I used the program BUCKy 1.4.4 [82, 83] to conduct a Bayesian concordance analysis, using the individual gene trees as input.
Taxon sampling: UCE dataset
Because the phylogeny inferred with Sanger sequencing data showed little support for deep relationships within the genus (see Additional file 7), I selected several exemplars of each of the well supported lineages for further analysis using ultra-conserved elements (UCEs). In total, I used 13 outgroup and 37 ingroup species in an effort to resolve the poorly supported backbone of the phylogeny.
Sequence generation: UCE dataset
I input up to 50 ng of DNA, sheared to a target fragment size of 400–600 bp into a genomic DNA library preparation protocol following Faircloth et al.  as modified by Branstetter et al. . I performed enrichments on pooled libraries using a set of 9898 custom designed probes (MYcroarray) targeting 2524 UCE loci in Formicidae. I followed the library enrichment procedures for the MYcroarray MYBaits kit, except that I reduced the standard MYBaits concentration to 0.1X, used custom adapter blockers instead of the standard MYcroarray blockers, and left enriched DNA bound to streptavidin beads during PCR, as described in Faircloth et al. . Following post-enrichment PCR, I purified the resulting pools using magnetic speedbeads and adjusted their volume to 22 μL.
I verified enrichment success and measured size-adjusted DNA concentrations of each pool with qPCR using a SYBR® FAST qPCR kit (Kapa Biosystems) and a Bio-Rad CFX96 (Bio-Rad Laboratories), and combined all pools into an equimolar final pool. The final pool was sent to the High Throughput Genomics Facility at the University of Utah for sequencing as a single lane on an Illumina HiSeq 2500 (125 cycle paired end sequencing v4).
Processing and alignment: UCE dataset
I demultiplexed the FASTQ data output, then trimmed away adapter contamination and low-quality bases using the program illumiprocessor, which is included in the PHYLUCE package. Cleaned reads were assembled de novo with PHYLUCE using a script that employs the program Trinity v 2013–02-25 . I mapped the contigs from the Trinity assembly to UCE loci and extracted FASTA sequences from the matched contigs. I aligned all loci individually using the L-INS-I algorithm in MAFFT. After alignment, I used ZORRO to estimate alignment confidence values for all loci, and trimmed all sites that fell below an arbitrary threshold of five.
Phylogenetic inference: UCE dataset
To test the sensitivity of tree topology and node support to missing data, partitioning scheme, compositional heterogeneity, evolutionary rates, gene tree-species tree conflict, and tree inference method, I ran a series of 25 analyses on 18 data subsets.
To examine the effect of missing data on phylogenetic inference, I constructed a series of matrices from loci that contained a minimum of 25%, 50%, 75%, 90%, 95%, 99% and 100% of all taxa (‘min25’, ‘min50’, ‘min75’, ‘min90’, ‘min95’, ‘min99’, ‘min100’ datasets). I then calculated summary statistics and estimated a phylogeny for each concatenated matrix with AMAS  and RAxML, respectively. For each RAxML tree inference, I used an unpartitioned matrix with 200 rapid bootstrap replicates, the GTR + G substitution model, and estimated the best tree annotated with bootstrap support values for each node. I wrote a custom shell script to calculate average bootstrap support for each tree. By comparing number of parsimony informative sites, proportion of parsimony informative sites, and average ML bootstrap support against percent missing data, I selected the 90% complete matrix (‘min90’) and the 75% complete matrix (‘min75’) for further analyses (see Additional file 8: Table S5 for summary statistics).
To select data partitions for the UCE data on the ‘min90’ matrix, I used the kmeans and rcluster clustering algorithms as implemented in the program PartitionFinder 2 [77, 88,89,90] to partition the ‘min90’ matrix into data subsets that share similarly evolving sites, limiting the model search to GTR + G. For the kmeans search, I used the complete concatenated dataset as input, for rcluster I used the full dataset partitioned by UCE locus. Additionally, I used a script developed by Borowiec  to extract protein coding sequence from the ‘min75’ dataset, then trimmed and aligned the extracted sequence using the local version of the program TranslatorX  (‘coding’ dataset). I inspected each alignment by eye using Aliview , and discarded sequences that were mismatched and most likely contaminants. I then partitioned the protein-coding data from each locus by codon position and used the rcluster algorithm in PartitionFinder 2 to gather these into similarly evolving subsets. I used the best partitioning schemes found by PartitionFinder to estimate trees with RAxML, using the same settings as in the missing data analysis above. To investigate the properties of single UCE loci in preparation for downstream analyses, I estimated single, unpartitioned UCE locus trees in RAxML using the settings listed above for the ‘min90’ dataset.
For BI analyses, I used the ‘min90’ dataset partitioned by the kmeans and rcluster algorithms, and the ‘coding’ dataset partitioned by rcluster as input. For each dataset, I used ExaBayes  to execute two independent runs, linking branch lengths across partitions and running each analysis for 1 million generations. I assessed convergence among runs and run performance with the program Tracer v1.6.0. Additionally, I inspected PSRF values, calculated average standard deviations of split frequencies, and constructed consensus trees using scripts included in the ExaBayes package.
Sensitivity analyses: UCE dataset
To inspect the sensitivity of tree topology and node support to subsets of UCE data, I conducted an array of analyses to examine the effects of compositional heterogeneity (‘rcfv’ datasets), evolutionary rate (‘slow’ datasets), random samples of UCE loci (‘rand’ datasets), and gene-tree species-tree conflict (‘ASTRAL’ analysis). See Additional file 2 for more details.
Four ‘backbone’ nodes were poorly supported among the UCE analyses (labeled i-iv in Fig. 2) and each was subjected to constraint analyses. Because Bayesian analyses are computationally expensive on large datasets, I performed the stepping stone sampling method developed by Xie et al.  on the full Sanger sequencing dataset, constraining the trees into two sets of competing hypotheses to calculate Bayes factors, in accordance with the recommendations of Bergsten et al.  (see Additional file 2 and Additional file 3: Table S1 for the hypotheses tested).
I ran these analyses in MrBayes 3.2.6 with two independent runs per analysis, setting the value of the alpha-shape parameter of the beta distribution to 0.4, number of chains to four, and sampling 25.5 million MCMC generations every 100th generation for 50 steps between the posterior and the prior, and discarding the initial 500 thousand MCMC steps as burn-in. After initial runs, the marginal likelihoods of the topology tests within Temnothorax were all very close; I subsequently increased the number of independent runs for each test at nodes i-iii to four.
I reconciled the results of these two sets of analyses by constraining the backbone of the more species-rich Sanger sequencing dataset based on the results of the UCE analyses and stepping stone analyses, and re-estimated the phylogeny in BI and ML frameworks (see Fig. 3 for the final constraints used). For BI, I used MrBayes with the settings for the full, unconstrained dataset analysis. For the ML analyses I used and a beta version of IQ-TREE and RAxML, with the settings used above in the unconstrained analyses, and using the ‘–g’ option for both programs, which uses a multifurcating tree as input for the constrained tree search.
To properly place the fossil taxon Temnothorax praecreolus in the phylogeny in preparation for downstream divergence dating analyses, I recorded characters and took measurements from this fossil, in addition to all Temnothorax worker specimens used in the phylogeny. The final dataset included 29 characters, which were either inherently discrete, discretized continuous characters, or discretized indices (see Additional file 9 and Additional file 10: Table S6).
Divergence time analysis
Divergence dates in Temnothorax were inferred using a combination of node- and tip-dating approaches in the program BEAST 2.3.0 . I used morphological data to place Temnothorax praecreolus, but due to the uncertainty of the placement of Baltic amber specimens (see Additional file 1), I compared two analyses, using these fossils to calibrate: (a) the node subtending the CF (b) the node subtending [LGG + Temnothorax] (see Additional file 2 for details on the calibrations used). I used the combined concatenated Sanger molecular and morphology dataset as input, partitioning the molecular data with the scheme suggested by ParititionFinder. I constrained the tree topologically based on the results of the UCE and stepping stone analyses (see Fig. 3 for constraints). Each analysis consisted of two runs of 100 million generations, sampling every 10 thousand generations. Time series plots and ESS values were analyzed with Tracer to assess burn-in and to determine whether runs had converged and chains were mixing adequately. I summarized the tree files with LogCombiner v1.8.0 and TreeAnnotator v1.8.0 after discarding the initial 25% of trees as burn-in.
I used the consensus chronogram from the BEAST analysis as input for the likelihood-based R package BioGeoBEARS , pruning all outgroup taxa from the tree except for the CF. I used a biogeographical classification scheme following previous studies (e.g. ), but discretized the Caribbean islands, including the Bahamas, into a separate biogeographical unit given the high diversity of Temnothorax species on these islands, and their historical isolation from the mainland  (see the key in Fig. 4 for a summary of the areas used). Because no extant species in the phylogeny inhabits more than two biogeographical areas, I limited the maximum ancestral species range to 2 units. Because the ancestor of the CF is inferred to have arisen between 60 and 50 Ma, relatively little tectonic drift with respect to the current positions of the continents has taken place in the evolutionary history of this group; therefore, I used a single time slice for this analysis. I imposed matrices of dispersal constraints, dispersal multipliers, area adjacencies, and allowed ancestral ranges on the analysis (see script and input files on Dryad). I ran the program over the six models included in the example script, and compared these via AICc score. Because the models offered alternate reconstructions, I averaged the results across all six models using the AICc weights in the output file. Finally, I examined the sensitivity of the ingroup reconstruction to incomplete outgroup sampling (see Additional file 2 for details).
Ancestral state reconstruction
To reconstruct the historical shifts in nesting habitat and parasitism in Temnothorax, as well as the potential effects of these characters on diversification rates, I estimated the ancestral states of these two traits using the hidden state speciation and extinction (HiSSE) model . I gathered nesting habitat and behavioral data from the literature, collection information, and personal observation (see Additional file 11: Table S7), and coded all extant taxa according to whether they are:
arboreal, nesting exclusively in vegetable matter, living or dead, above the soil and leaf litter layer, or
terrestrial, nesting in variety of habitats, but always including the soil, leaf litter layer, and/or fallen, rotten logs.
In the parasitism analysis, species were coded as:
free living, if nests are founded independently, or
social parasites, if new nests are founded exclusively in the nests of other ants; I include the xenobiotic genus Formicoxenus in this category.
Because the taxon sampling in this phylogeny is incomplete, I used global sampling fractions for each analysis, modeled as the proportion of species in each state that are included in the phylogeny.
For the nesting habit dataset, I searched among 59 models, composed of BiSSE-like models, variations on the HiSSE model, and null models in which diversification rates are independent of character states (see Additional file 2 and Additional file 12: Table S8 for details on model specification). I included transition-restricted models that disallowed transitions from arboreal to terrestrial nesting and, based on preliminary results, models that disallowed a hidden state within arboreal nesting. HiSSE initially returned a surprising result, showing that a hidden state associated with arboreal nesting was responsible for a two-fold decrease in net diversification rate. I inspected reconstructions of the states on the tree, and found a single transition to the hidden state at the root of the tree, subtending the long-branched outgroups. After trimming Gauromyrmex and Vombisidris from the dataset, all traces of character dependent diversification rates disappeared.
For the parasitism dataset, I constructed 41 models using most of the models mentioned above, with a subset disallowing transitions from parasitism to free living, but unlike the nesting habit model search, I didn’t include HiSSE models that disallowed hidden states. For both datasets I summarized all reconstructions, accounting for model uncertainty, by averaging over all models after weighting each model by its AIC score.
Sanger dataset characteristics
The resulting concatenated data matrix was 8010 bp in length with 4.1% missing data, containing 3469 (43.3%) variable sites and 2608 (32.5%) parsimony-informative sites, with most of the variability (52%) concentrated in the third codon positions of the alignment (see Additional file 13: Table S9). The data matrix is nearly complete, with COI + COII missing for the outgroups and six ingroup taxa (Temnothorax dr02, T. mpala, T. rothneyi, T. smithi, T. cf. striatulus, and T. unifasciatus), the first exon of Wg missing for several outgroup taxa (Formicoxenus diversipilosus, Harpagoxenus sublaevis, Vombisidris bilongrudi, and Tetheamyrma subspongia) and ingroup taxa (Temnothorax mmp15, T. mpala, T. obturator, T. rothneyi, T. semiruber, and T. smithi). In Temnothorax stenotyle, the intron of ArgK has apparently been lost: both exonic fragments amplified normally, but the intervening intron was missing upon assembly and alignment. I confirmed this by inspecting an argarose gel of the amplicon that usually contains intron; the fragment was shorter by approximately 100 bp in comparison to other taxa.
Among the nuclear protein coding genes, the exons exhibit between 22 and 48% parsimony informative sites (PIS), with COI + COII having the greatest proportion, and abdA having the smallest. The exonic data contributed the most PIS to the Sanger dataset, containing 2070 sites out of 2608 in the entire matrix, or 79% of all informative sites. Introns contained a high proportion of informative sites as well, with 29–56% PIS, contributing 424 informative sites the overall dataset (16.5% of the total PIS). 28S was the most conserved DNA fragment used in the dataset, contributing only 114 informative sites, or 4.5% of all PIS in the dataset.
Inspection of base frequencies among the data subsets revealed that the third codon position of COI + COII, as well as the introns of all fragments except LWRh are highly AT biased, with the proportions ranging between 72 and 80%. Conversely, AbdA and the third codon positions of all nuclear genes, save CAD, were GC rich, with proportions ranging between 60 and 77%. Aside from the third codon position of COI + COII, no subset of the data displayed significant compositional heterogeneity among taxa.
Sanger dataset phylogeny
The best scheme found by PartitionFinder consisted of 14 partitions (see Additional file 14: Table S10). The best tree with branch lengths from the BI analysis of the unconstrained, concatenated and partitioned data is shown in Additional file 7, Fig. H, including support values from both Bayesian and maximum likelihood analyses. The trees from the BI and ML analyses are nearly identical in topology, differing in inferred relationships and branch lengths only at nodes that were weakly supported in both. Both methods recovered Temnothorax as a monophyletic group with high support (PP = 1, BS = 100). In concordance with Ward et al. , the Indomalayan Gauromyrmex my01 was recovered as sister to all other CF in the study, and the LGG was monophyletic across analyses. In contrast to the relationship found in Ward et al. , the Australasian species Vombisidris bilongrudi was inferred to be sister to Temnothorax, but this relationship received only moderate support (PP =1, BS = 53–71). Furthermore, ten well-supported clades were identified within Temnothorax, although the relationships among these clades were poorly supported in all analyses based on Sanger data. All ten clades had maximum support (PP =1 and BS =100), except for clade 2 (PP 1, BS 88) and clade 5 (PP 1, BS 99). Within the ‘Palearctic’ clade, three well-supported subclades emerged (PP 1, BS 100), in addition to a fourth subclade with mediocre support (PP 1, BS 63).
In general, where trees resulting from data exclusion experiments differed topologically from analyses using the full dataset, it was almost exclusively at nodes that were weakly supported across analyses. Many of the relationships within the salvini clade continued to have low support among data subset analyses, with several highly-supported species pairs and one well-supported subclade. The position of the mmp11 group was extremely unstable among analyses, being placed variously as sister to the rottenbergi clade, as sister to [sallei subclade + iris subclade + tricarinatus group], or as sister to all Temnothorax. Additionally, the topology within the Palearctic clade was very plastic among analyses, but was consistently composed of the above mentioned four subclades, in addition to the rogue taxa T. nylanderi and T. oxianus.
The primary concordance tree with branch lengths in units of concordance factors is given in Additional file 7. Concordance factors are interpreted as the estimated proportion of the data that supports a given node, for example, the node subtending Temnothorax is supported by a high concordance factor (CF = 0.88). While all major subclades were recovered (except for the mmp11 group and the gunlachi group), only five of the major clades had concordance factors over 0.4. The four subclades of clade 10 were recovered as well. Again, the relationships among all the 10 major clades of Temnothorax were poorly supported, with the concordance factors of the “backbone” of the phylogeny ranging between 0 and 0.18. After considering single gene consensus trees (see Additional file 15), this is most probably due to a paucity of informative sites in the dataset that resolve the backbone of the phylogeny, rather than gene-tree conflict.
UCE dataset characteristics
After assembling the data and matching probes to contigs, the mean number of UCE loci per specimen was 2314, with a mean length of 1002 bp and a mean coverage score of 60 X (see Additional file 16: Table S11). After filtering the loci by weighing matrix completeness against proportion of parsimony informative sites and mean bootstrap support, I selected the ‘min90’ dataset for further analysis. The ‘min90’ dataset included 2098 UCE loci with a mean length of 744 bp, which resulted in an aligned, concatenated matrix 1,561,581 bp in length, of which 404,463 sites (26%) were parsimony informative. All sequences have been submitted to the NCBI Sequence Reads Archive (BioProject PRJNA393044).
The dataset was slightly AT rich, with these bases comprising 56% of the dataset. In a chi-squared test, no loci were found to be diverging significantly from base composition homogeneity across taxa.
UCE dataset phylogeny
A cladogram with the topology from the ExaBayes analysis with the ‘rcluster’ partitioning scheme is contrasted with the phylogram from the same analysis and presented in Fig. 2; see Additional file 17 for a synopsis of all trees resulting from filtering treatments. The ML and BI approaches using the ‘rcluster’ partitioning scheme recovered identical topologies, differing only in their support of certain nodes (PP = 1, BS ≥ 90). Across the sensitivity analyses, however, topology diverged at several nodes. The phylogenetic position of Vombisidris bilongrudi was particularly volatile: it was found to be sister to the clade consisting of the LGG + Temnothorax in the ‘slow33’, ‘ML kmeans’, ‘BI kmeans’, and ASTRAL analyses, while it was found to be sister to Gauromyrmex + the LGG + Temnothorax in the coding data only, and two of the ‘rand33’ and the ‘rand66’ datasets each.
Another node that was unstable across a large subset of analyses was the node subtending the iris subclade and tricarinatus group; in this case, the trees resulting from the ASTRAL, two of the rand33 datasets, and the coding data only datasets recovered the iris subclade as sister to the mmp11 group with low-to-moderate support (BS = 47–85). The ‘slow33’ and one of the ‘rand33’ loci datasets recovered the iris subclade as sister to [tricarinatus group + mmp11 group] with high support (BS = 96–100).
The positions of the rugatulus and rottenbergi clades were unstable among analyses as well, with the rugatulus clade being recovered as sister to [salvini group + obturator group + Palearctic clade] in the ‘min95’ and coding data only datasets. In the ‘min99’ dataset, the obturator clade was recovered as sister to the Palearctic clade. Within the salvini clade, the ‘slow33’ dataset recovered T. pergandei as being sister to T. cf. fuscatus + T. subditivus. In the BI coding data only analysis, T. recedens formed a clade with T. gredosi. It is worth noting that in the case of Vombisidris and T. gundlachi, the datasets composed of a reduced set of random loci often arrived at the same topology as the other reduced datasets, suggesting that in these cases, the topology may rest on a small fulcrum of data.
The subclades of Temnothorax that were identified in the analysis of the Sanger sequencing dataset are recovered with maximum support across all analyses, except for the iris subclade, which is only represented by T. gundlachi in the UCE dataset. Other than the nodes highlighted above, the relationships among the ten subclades that were identified in the Sanger analysis are fully resolved, receiving maximum support across all analyses. Additionally, the rogue taxa T. oxianus and T. nylanderi are recovered as members of the subclades ‘I’ and ‘III’ of the Palearctic clade, respectively.
The results of the constraint analyses are given in Table 1. I interpret the results of the Bayes factor tests following the suggestions of Kass & Raftery , assigning levels of support to the following ranges of 2lnBF: 0–2, not worth explaining; 2–6, positive support; 6–10, strong support; and >10, very strong support.
node i. rugatulus clade sister to [obturator clade + Palearctic clade + salvini clade] received strong support 2lnBF = 6.96).
node ii. obturator clade being sister to the Palearctic clade vs. the salvini clade was equivocal (2lnBF = 0.62).
node iii. Iris subclade sister to the mmp11 group received strong support (2lnBF = 6.94).
node iv. Vombisidris as sister to the LGG + Temnothorax was positively supported (2lnBF = 5.76).
The chronogram from the BEAST analysis of the constrained Sanger sequencing and morphology dataset is presented in Fig. 4. Temnothorax praecreolus, (not shown) was recovered as sister to a clade of mainland Meso-American species of the salvini clade in the BEAST analysis, whereas in the MrBayes analysis it was sister to a clade of mostly Caribbean salvini clade species, the pulchellus species-group (see Additional files 18 and 19). The BEAST chronogram differs topologically from the MrBayes constrained phylogram (see Fig. 3) at several nodes with low support: in the rottenbergi clade, T. cristinae and T. cf. flavispinus switch positions; the positions the non-pulchellus group members of the salvini clade continue to be unstable, as in the constrained BI and ML analyses; the positions of all species within the Palearctic III subclade except T. nylanderi and T. congruus are unstable, as well as all non-flavicornis, muellerianus and laurae group species in Palearctic IV subclade.
The crown group of Temnothorax is estimated to have arisen 35 Ma (95% highest posterior density (HPD) = 28–41 Ma), with the stem group + LGG emerging roughly 11 Ma earlier: 46 Ma (95% HPD = 39–54 Ma). Immediately following the origin of the crown Temnothorax at the Eocene-Oligocene transition, there appears to have been a rapid succession of splitting events. Interestingly, the LGG, and Temnothorax to a lesser extent, are both subtended by long branches (38 Ma in the LGG, 11 Ma in Temnothorax). Similarly, long branches subtending large clades (> 5 Ma) appear in the rottenbergi-, rugatulus- and obturator clades, suggesting extensive pruning if speciation is assumed to be constant over time.
The BioGeoBEARS analysis of the empirical data identified the dispersal-extinction-cladogenesis + jump dispersal (DEC + J) model as the best fit in comparison to its null model, DEC, and the four other models tested (see Additional file 20: Table S12). The jump (+J) models always performed significantly better than the alternatives, and the results of the models with the second and third highest likelihoods, DIVALIKE + J and BAYAREALIKE + J, largely agree with the DEC + J model, differing mainly in the size of ancestral areas at several nodes (see [Additional file 21: Table S13). The model-averaged results of the empirical data agreed with the results of the sensitivity analysis almost exactly, except for the LGG, which was reconstructed as having a Holarctic MRCA in the sensitivity + empirical average (see Additional file 21: Table S13 and Additional file 22).
The results suggest an ancestral Palearctic-Indomalayan distribution for the origin of the CF (node 1), with Gauromyrmex and Vombisidris remaining in the Indomalayan realm, with the latter genus presently extending into the Australasian region (node 2). There is some uncertainty surrounding this reconstruction though, with alternative empirical models and sensitivity analyses suggesting a strictly Indomalayan origin, or an Indomalayan Australasian origin (see Additional file 21: Table S13). Among nearly all models and sensitivity analyses, the ancestral range of the LGG + Temnothorax is estimated to be Palearctic (node 3). There is quite a bit of uncertainty about the ancestral range of the LGG, with most of the sensitivity analyses suggesting a Holarctic origin, while the empirical data suggests a Palearctic range (node 23). The MRCA of Temnothorax, on the other hand, was reconstructed as Nearctic across nearly all the analyses (node 4).
The ancestor of all crown Temnothorax dispersed to the Nearctic sometime in the middle to late Eocene, 35–47 Ma (nodes 3, 4). Much of the early diversification of Temnothorax is inferred to be in the Nearctic realm, with a striking pattern of dispersal: soon after a rapid series of splitting events, there was a period of large-scale dispersal, spanning the Oligocene to middle Miocene. During this period, at least four separate lineages expanded their range or dispersed into the Greater Antilles. The ancestor of the sallei subclade is inferred to have had a Nearctic-Caribbean distribution (node 22), retained in the extant species T. allardycei, which has variously been categorized as an exotic species from the Neotropics [102, 103], or a species endemic to the Bahamas and Florida [104,105,106]. A second dispersal event in the sallei clade occurred in the same time frame, represented by the terricolous species T. gundlachi and T. poeyi (node 21). The ancestor of two members of the obturator clade, T. creolus and T. mmp15, the former found on the island of Hispaniola and the latter an endemic of Puerto Rico, also dispersed to the Caribbean from the Nearctic in the early Miocene (nodes 14, 15). Several members of the pulchellus group in the salvini clade also dispersed to the Caribbean islands from the mainland Neotropics in the early Miocene (nodes 12, 13), with T. torrei extending its range to the Nearctic sub-tropics between the Pliocene and the present.
Another series of Oligocene to middle Miocene dispersals to the mainland Neotropics from the Nearctic occurred in several disparate lineages: the salvini clade has an extensive history of diversification in this realm, dispersing from the Nearctic in the early Oligocene (nodes 6, 11) with the ancestors of T. pergandei, T. peninsularis, and T. mmp12 dispersing back to the Nearctic sometime between the middle Miocene and the present. In the sallei clade, the ancestor of the mmp11 group and the iris subclade dispersed to the Neotropics at some point during the early Oligocene, 27–31 Ma (nodes 19, 20), with the iris subclade dispersing to the Caribbean 13–27 Ma (nodes 20, 21); mmp08 dispersed much later, at most 12 Ma, while T. striatulus extended its range into the Neotropics during this period. A single lineage from the obturator clade is known from cloud forest in the Neotropical region, dispersing from the Nearctic at some point after the early Miocene.
Perhaps the most significant dispersal events during the Oligocene to middle Miocene involve movements between the Nearctic and Palearctic. For example, the ancestor of the extremely diverse Palearctic clade dispersed to Eurasia from the Nearctic between 27 and 31 Ma (nodes 6, 7), the ancestor of T. clypeatus and T. corticalis dispersed between 16 and 23 Ma nodes 16, 17), and the ancestor of the rottenbergi clade dispersed between 23 and 33 Ma (nodes 5, 18). Because the dates for these dispersal events were well after the Eocene, I rule out the early Eocene trans-Atlantic Thulean route ; presumably these occurred over the Beringian land bridge which was present intermittently throughout the Paleocene to the late Miocene . A hypothesis of east to west movement to the Palearctic out of the Nearctic is supported with a pectinate series of lineages in the Palearctic clade: the Palearctic II subclade, composed of mostly east Asian taxa, is sister to the mostly central and western Eurasian Palearctic III clade and the mostly Mediterranean Palearctic IV subclade. The Palearctic I subclade, however, is a bit enigmatic in that T. oxianus, collected from Kyrgyzstan, is the sister taxon to a clade of western European and eastern North American taxa, suggesting a back-dispersal to the Nearctic (nodes 9, 10). Within the Paleartic IV subclade, dispersal to the Afrotropics from the Palearctic occurred after the late Miocene (node 8), with the split between T. mpala and T. at01 closely coinciding with the most recent estimates of Saharan aridification, which occurred 7–11 Ma .
Ancestral state estimation
After trimming the long-branched outgroups from the nesting habit dataset, there were two models that fit equally well based on AIC score. Both were relatively simple HiSSE models with three free parameters, in which each had one turnover rate, one extinction fraction for all states, transition rates equal, and no hidden state within arboreal nesting. They differed in that one model allowed transitions from arboreal to terrestrial nesting, and one didn’t. No character dependent diversification was detected for arboreal nesting. The model-averaged ancestral state reconstruction suggests that the MRCA of Temnothorax was ground-nesting, with multiple independent transitions to arboreal nesting in the Oligocene (see Fig. 4). Transitions from arboreal to terrestrial nesting were recovered in only one case, in the rugatulus clade.
The best model for the parasitism dataset was a complex BiSSE-like model with all parameters free. However, several competing models fell within a ΔAIC of 2 or less, and among these was a character independent diversification model (CID-4, a four-rate model with transition rates equal). Indeed, when the models are averaged, there is little support for character dependent diversification when parasitism is considered (see Additional file 23). The model-averaged ancestral state reconstruction suggested multiple independent transitions to parasitism, all of which were in the middle Miocene or later.
The phylogenetic analyses from two sequencing approaches and on multiple data subsets recover identical clades in Temnothorax and largely agree on the relationships among them. Although even the large UCE dataset could not resolve some of the relationships in the backbone satisfactorily, constraint analyses on the Sanger dataset resolved most of the difficult nodes. Temnothorax was recovered as a natural group with high support, as was the sister group relationship the Leptothorax genus group. The species groups, former subgenera, and satellite genera of the core formicoxenines, including Temnothorax have a long and controversial history. A phylogenetic framework for understanding the relationships within this intriguing group of ants has been sorely lacking, which has hampered taxonomic revisionary efforts and our understanding of the pace and mode of evolution within this genus. In this study, the inclusion of the former genera Chalepoxenus, Myrmoxenus, and Protomognathus in Temnothorax is found to be justified. The former genus Macromischa is revealed to be polyphyletic, confirming earlier suppositions [12, 41], and composed of members of the salvini, obturator, and sallei clades, interdigitated with the former subgenus Myrafant. A striking pattern within Temnothorax is strong morphological conservatism within and between groups (e.g., compare taxa from the tricarinatus group and the andrei, rugatulus, and Palearctic clades) punctuated by extreme morphological departures (e.g. the Caribbean members of the sallei clade, and the salvini clade).
The Cenozoic has been a period of dramatic fluctuations in climate that have substantially affected faunal and floral distributions . The diversification of the CF encompasses much of this era, so any discussion of evolution needs to be viewed in the context of responses to changes in climate regime. Furthermore, the historical biogeographic patterns of Holarctic flora and fauna are some of the most thoroughly studied, and are generally understood as the products of a continuous habitat across the Northern Hemisphere, connected intermittently by the De Geer and Thulean routes across the North Atlantic during the late Cretaceous and early Tertiary [109, 110], or nearly permanently by the Beringian route from the middle Cretaceous through the Tertiary, and then intermittently connected during the Quaternary [110,111,112]. In the discussion below, I present the historical biogeography and trait evolution of Temnothorax, considering a timeline of major geologic events during the evolution of the genus.
The origins of the CF are obscured by uncertainty in the divergence dates, but the biogeographic analysis suggest that the MRCA of this clade arose 47–64 Ma with a Palearctic-Indomalayan range. This period includes late Paleocene Thermal Maximum (LPTM) , an event that was marked by a steep 5–6 °C rise in deep-sea temperature over a very short time period, < 10 ka (thousand years) [113,114,115], and the wide-scale dispersal and radiation of land plants and mammals [116,117,118,119,120], likely facilitated by expansion of broadleaf evergreen forests into high latitudes [121,122,123]. By 52 Ma, at the Eocene Climatic Optimum, the ancestor of the LGG + Temnothorax had a more restricted Palearctic distribution. The ancestor of Temnothorax dispersed to the Nearctic during the middle to late Eocene, but there is little evidence to suggest whether this was by either the Thulean or Beringian route.
The results of this study support an origin of the crown-group Temnothorax in the Nearctic realm roughly 35 Ma, at the Eocene-Oligocene transition, running contrary to Bernard’s ‘out-of-the-Palearctic’ hypothesis. Early rapid diversification in the Nearctic was followed by large-scale dispersal. By this time, the North Atlantic Thulean route had closed, leaving Beringia as the most probable dispersal route (although Tiffney  suggests that a North Atlantic route may have operated as a series of stepping stone islands into the middle Tertiary). The Eocene-Oligocene transition (34 Ma) is characterized by another drastic swing in global temperature, this time with the deep-sea cooling rapidly by 5 °C, initiating the formation of an Antarctic icecap and a glacial period that lasted roughly 400 ka. Like the LPTM, this event had dramatic effects on the global biota, leading to abrupt turnover [124, 125], the origination of many plant and animal lineages [126, 127], retraction of evergreen broadleaf forests toward the equator, expansion of seasonal deciduous forest, and widespread aridification [121, 122]. Given the contemporary habitat affinities of many species in the genus, the common ancestor of Temnothorax may have already adapted to living in cool and/or xeric habitats, and thus could fully exploit these habitats as they expanded during the Eocene-Oligocene transition. This also may have facilitated movement across Beringia of the rottenbergi and Palearctic clades, a region which, following the transition, was dominated by mixed mesophilic forest . Movement of the ancestor to the salvini clade into the Neotropics also occurred during the early Oligocene, where many of the contemporary taxa inhabit mesophilic forests, although some taxa are found primarily in dry forest or littoral scrub (pers. obs.).
Temnothorax appears to be one of several speciose Northern Hemisphere social insect lineages that originated or radiated in association with the Eocene-Oligocene transition: these include the cold-adapted Bombus bumblebees , Myrmica ants  and Stenamma ants . In each of these cases, the biogeographic history suggests a great fluidity of dispersal across Beringia, with 21 clades of Bombus exploiting this route starting in the Miocene (~20 Ma), several dispersal events in the case of Myrmica from the Oligocene onwards, and three Beringian dispersals in Stenamma, starting in the late Eocene. In the case of Bombus, dispersal is asymmetric, with Palearctic lineages moving to the Nearctic at higher rates; in Stenamma, two Nearctic-to-Palearctic events are inferred and one Palearctic-to-Nearctic; in the case of Myrmica, no formal biogeographic analysis was performed but it appears that dispersal occurred in both directions. Most of the dispersal in Temnothorax appears to be in the Nearctic-to-Palearctic direction, with one case of dispersal back to the Neactic occurring in the early Miocene, ~17 Ma in the case of the ancestor of T. americanus and the longispinosus group.
Global temperatures remained relatively stable until the late Oligocene, when a brief period of warming (ca. 24 Ma) was followed by another steep drop in deep ocean temperature, initiating a second period of glaciation at the Oligocene-Miocene boundary (23 Ma) , aridification [131, 132], and faunal turnover in at least some lineages . Several lineages of arboreal Temnothorax arose near this transition, notably the Caribbean sallei subclade, the obturator clade, and members of the salvini clade, and continued to diversify throughout the Miocene. The causes behind this concerted pattern are obscure; epiphytic habits evolved repeatedly in the bromeliads during this era , and several Neotropical salvini clade species are associated with the genus Tillandsia . Or, perhaps the abrupt drop in global temperatures opened a niche that was previously filled by a less cold-tolerant taxon.
Glaciation events continued throughout the Miocene, but intermittently and on a smaller scale than the event that marked the Oligocene-Miocene boundary . The early Miocene was a period of generally warm temperatures, which slowly fell starting in the middle Miocene, when the Antarctic ice-sheet became a persistent global feature. Multiple lineages dispersed to the Caribbean during the Miocene, including the ancestors of the sallei and iris sublclades, the pulchellus group, and members of the obturator clade. These events all occurred in the early Oligocene to middle Miocene, 18–30 Ma, and in most of these lineages the ancestral mainland range is inferred to be Nearctic, with the exception of the pulchellus group which had a Neotropical ancestral range. Because the evidence suggests that the Greater Antilles have not been connected to mainland North America since the early Eocene , and because by the Greater Antilles + Aves Ridge (GAARlandia) land span that connected northern South America to the developing northern Greater Antilles had subsided by the late Oligocene , an over-water route is the most plausible method of dispersal for all lineages present on the Caribbean islands today. Caribbean paleoceanographic hypotheses presented by Iturralde-Vinent  dovetail with an over-water dispersal route: there appears to have been a counterclockwise loop current in the Gulf of Mexico drawing water down from the Florida peninsula to the Greater Antilles, which has essentially been continually operating from the late Eocene up to the present. Dispersal has been in a single direction, from mainland to islands, but one species has expanded its range back to the subtropical Nearctic in recent times (T. torrei).
There also appears to have been fluidity across the Nearctic-Neotropical realms during the Miocene. In the salvini clade, T. pergandei, T. sp. nr. peninsularis, T. mmp12, and T. subditivus have traversed this boundary independently, dispersing from the Neotropics to the Nearctic. An interesting observation is that many of these species inhabit, or are tolerant of, dry forest or littoral scrub habitats, which may hold the key to understanding their apparent mobility across the Neotropic-Nearctic realms. In the sallei and obturator clades, T. mmp11, T. cf. striatulus, T. striatulus, and T. mmp08 currently inhabit mesophilic forests in the Neotropics, from mid-to-high elevations. Temnothorax mmp08 is only known from the Sierra Morena in Chiapas, Mexico, while the other species are more widespread, ranging from the Sierra Madre del Sur in southern Mexico to Honduras. Given the habitat affinities of these taxa, it may be that these represent relictual sallei clade taxa, protected from episodic cooling and warming since the middle Miocene by the complex topography of Mesoamerica.
Ancestral state estimation
The ancestor of the crown Temnothorax is inferred to have had terrestrial nesting behavior, supporting the Bernard  hypothesis. The coordinated transitions to arboreal nesting behavior in several clades in the Oligocene are striking, but the mechanism driving this pattern remains unclear. Perhaps the late Oligocene warming facilitated movement into arboreal habitats; alternatively, early Miocene cooling may have increased competition among species for terrestrial nesting sites, with the more cold-tolerant Temnothorax species resorting to arboreal habitats. Within the ants, several genera, e.g. Camponotus, Crematogaster, Monomorium contain arboreal and ground-nesting species, but the history of transitions between these states has not been formally estimated in any of these. In Crematogaster, though, it appears that the predominant direction of transition is the converse of Temnothorax, with the predominantly arboreal and tropical species transitioning to terrestrial nesting in the temperate zones .
In this study, parasitism is inferred to have evolved several times within Temnothorax, corroborating the findings of Beibl et al. . Moreover, all of the parasitic taxa and hosts are confined to the Palearctic clade: the parasites are in subclades I and IV, while the hosts are scattered throughout subclades I, III, and IV (see Fig. 5). There appears to be support for both forms of Emery’s rule, but caution is needed when interpreting these results. Temnothorax americanus is the sister taxon to its hosts in the longispinosus group, but it is difficult to determine whether this is due to the ‘strict’ or ‘loose’ forms of Emery’s rule because a) the parasitic sister taxon is monospecific and b) the host species all have similar nesting habitats, typically within hollow twigs or nutshells in the leaf litter. Other species within the range of T. americanus are either parasites themselves (T. duloticus, T. minutissimus, and T. pilagens), have different ecological requirements (T. schaumii is arboreal, and T. texanus nests directly in the soil), or are distantly related (T. pergandei). Interestingly, T. tuscaloosae is not known to be parasitized by T. americanus, despite a range overlap in North Carolina.
T. bernardi and T. ravouxi of the corsicus group (formerly genus Myrmoxenus) are shown to be distantly related to their hosts in Palearctic subclades III and IV, suggesting the ‘loose form’ holds true in this case. Similarly, the muellerianus group (formerly genus Chalepoxenus) is the sister taxon to the laurae group; the parasitic muellerianus group taxa included here (T. muellerianus and T. kutteri) vary broadly with respect to their host species; but in all cases, they do not adhere to the ‘strict form’ of Emery’s rule. Recent phylogenetic studies of the muellerianus and corsicus species groups have elucidated the within-group relationships of these multi-species socially parasitic taxa . Although more comprehensive sampling would be desirable to infer the crown ages of these groups, the current study includes the most distantly related muellerianus group taxa from Beibl et al. . The results of the BEAST analysis suggest that these groups are young, with median ages of 4.1 and 4.4 Ma, respectively (see Additional file 24: Table S14).
This study represents the first effort to establish a time-calibrated phylogenetic framework for Temnothorax, elucidating a rich and intriguing history of evolution. Previous subgeneric and many species-group classification schemes were found to be unfounded with taxa distributed broadly among several morphologically convergent clades. Temnothorax arose in the Nearctic at the Eocene-Oligocene transition, ~35 Ma. The early evolution of the genus is distinguished by rapid radiation in a pectinate series followed by large-scale dispersal, with many of the major clades attaining their current geographic distributions by the early Miocene. Many arboreal lineages have a deep history dating back to the Oligocene, but they continued to arise in some clades throughout the Miocene. The ‘strict form’ of Emery’s rule for social parasites receives little support in the current study, although this may be case for T. americanus and other social parasites in the Palearctic I clade, namely T. minutissimus, T. duloticus, and T. pilagens. More thorough taxon sampling would further clarify the history of these interesting interactions. The new evolutionary framework presented here should simplify this task, having identified the phylogenetic context for the parasitic taxa and their hosts. Moreover, the evolutionary history of the Neotropical taxa is an exciting new frontier opened by this study. The convergence of two distantly related clades on the Macromischa syndrome and the radiation of Temnothorax on the Greater Antilles are intriguing features that beg further investigation.
Bayes factor support
Binary state speciation and extinction
Nucleotide base pairs
Maximum likelihood bootstrap support
Core formicoxenines, sensu Ward et al. 2015 , consisting of the genera Formicoxenus, Gauromyrmex, Harpagoxenus, Leptothorax, Temnothorax, and Vombisidris.
Hidden state speciation and extinction 
Highest posterior density
Leptothorax genus group, consisting of the genera Formicoxenus, Harpagoxenus, and Leptothorax.
Late Paleocene Thermal Maximum 
Million years ago
Most recent common ancestor
Parsimony informative sites
Bayesian posterior probability
Relative composition frequency variability
Hölldobler B, Wilson EO. The ants. Cambridge: Harvard University Press; 1990.
Ward PS, Brady SG, Fisher BL, Schultz TR. The evolution of myrmicine ants: phylogeny and biogeography of a hyperdiverse ant clade (hymenoptera: Formicidae). Syst Entomol. 2015;40:61–81.
Antwiki. Available from http://www.antwiki.org. Accessed 18 October 2017.
Bolton B. An online catalog of the ants of the world. Available from http://antcat.org. Accessed 18 October 2017.
Charbonneau D, Hillis N, Dornhaus A. ‘Lazy’ in nature: ant colony time budgets show high ‘inactivity’ in the field as well as in the lab. Insect Soc. 2015;62:31–5.
Achenbach A, Foitzik S. First evidence for slave rebellion: enslaved ant workers systematically kill the brood of their social parasite Protomognathus americanus. Evolution. 2009;63:1068–75.
Blatrix R, Herbers JM. Coevolution between slave-making ants and their hosts: host specificity and geographical variation. Mol Ecol. 2003;12:2809–16.
Heinze J, Rueppell O. The frequency of multi-queen colonies increases with altitude in a Nearctic ant. Ecol Entomol. 2014;39:527–9.
Möglich M, Maschwitz U, Hölldobler B. Tandem calling: a new kind of signal in ant communication. Science. 1974;186:1046–7.
Buschinger A. Sympatric speciation and radiative evolution of socially parasitic ants- heretic hypotheses and their factual background. J Zool Syst Evol Res. 1990;28:241–60.
Prebus M. Palearctic elements in the old world tropics: a taxonomic revision of the ant genus Temnothorax Mayr (hymenoptera, Formicidae) for the Afrotropical biogeographical region. ZooKeys. 2015;483:23–57.
Baroni UC. Materiali per una revisione dei Leptothorax neotropicali appartenenti al sottogenere Macromischa Roger, n. comb. (Hymenoptera: Formicidae). Ent Basiliensia. 1978;3:395–618.
Fontenla JL. New species of Leptothorax (hymenoptera: Formicidae) from Cuba. Avicennia. 1998;8/9:61–8.
Wheeler WM, Mann WM. The ants of Haiti. Bull Am Mus Nat Hist. 1914;33:1–61.
Fellers GM, Fellers JH. Scavenging rates of invertebrates in an eastern deciduous forest. Am Midl Nat. 1982;107:389–92.
Smith MR. An annotated list of the ants of Mississippi (Hym.). Entomol News. 1924;35:47–54.
Dennis CA. The distribution of ant species in Tennessee with reference to ecological factors. Ann Entomol Soc Am. 1938;31:267–308.
Douglas MM. Defense of bracken fern by arthropods attracted to axillary nectaries. Psyche. 1983;90:313–20.
Espadaler X. Leptothorax caesari sp. n.(Insecta: hymenoptera: Formicidae), a granivore with apterous males. Ann Nat Hist Mus Wien Ser B Bot Zool. 1997:145–50.
Fokuhl G, Heinze J, Poschlod P. Myrmechory by small ants – beneficial effects through elaiosome nutrition and seed dispersal. Acta Oecol. 2012;38:71–6.
Möglich M. Social organization of nest emigration in Leptothorax (Hym., form.). Insect Soc. 1978;25:205–26.
Möglich M. Tandem calling pheromone in the genus Leptothorax (hymenoptera: Formicidae): behavioral analysis of specificity. J Chem Ecol. 1979;5:35–52.
Beckers R, Goss S, Deneubourg J, Pasteels JM. Colony size, communication and ant foraging strategy. Psyche. 1989;96:239–56.
Alloway TM, Buschinger A, Talbot M, Stuart R, Thomas C. Polygyny and polydomy in three north American species of the ant genus Leptothorax Mayr (hymenoptera: Formicidae). Psyche. 1983;89:249–74.
Foitzik S, Heinze J. Microgeographic genetic structure and intraspecific parasitism in the ant Leptothorax nylanderi. Ecol Entomol. 2001;26:449–56.
Partridge LW, Partridge KA, Franks NR. Field survey of a monogynous leptothoracine ant (hymenoptera, Formicide): evidence of seasonal polydomy? Insect Soc. 1997;44:75–83.
Bourke AF, Heinze J. The ecology of communal breeding: the case of multiple-queen leptothoracine ants. Phil Trans R Soc B. 1994;345:359–72.
Frumhoff PC, Ward PS. Individual-level selection, colony-level selection, and the association between polygyny and worker monomorphism in ants. Am Nat. 1992;139:559–90.
Guénard B, Shik JZ, Booher D, Lubertazzi D, Alpert G. Extreme polygyny in the previously unstudied subtropical ant Temnothorax tuscaloosae with implications for the biogeographic study of the evolution of polygyny. Insect Soc. 2016;63:543–51.
Herbers JM. Effects of ecological parameters on queen number in Leptothorax longispinosus (hymenoptera; Formicidae). J Kans Entomol Soc. 1986;59:675–86.
MacKay WP. A review of the new world ants of the subgenus Myrafant (genus Leptothorax) (hymenoptera: Formicidae). Sociobiology. 2000;2000:36.
Seifert B. Die Ameisen Mittel-und Nordeuropas. Tauer: Lutra Verlags-und Vertriebsgesellschaft; 2007.
Wilson EO. The insect societies. Cambridge: Belknap Press of Harvard University Press; 1971.
Moreau CS, Bell CD, Vila R, Archibald SB, Pierce NE. Phylogeny of the ants: diversification in the age of angiosperms. Science. 2006;312:101–4.
Brady SG, Schultz TR, Fisher BL, Ward PS. Evaluating alternative hypotheses for the early evolution and diversification of ants. Proc Natl Acad Sci U S A. 2006;103:18172–7.
Buschinger A. Social parasitism among ants: a review (hymenoptera: Formicidae). Myrmecol News. 2009;12:219–35.
Beibl J, Stuart RJ, Heinze J, Foitzik S. Six origins of slavery in formicoxenine ants. Insect Soc. 2005;52:291–7.
Beibl J, Buschinger A, Foitzik S, Heinze J. Phylogeny and phylogeography of the Mediterranean species of the parasitic ant genus Chalepoxenus and its Temnothorax hosts. Insect Soc. 2007;54:189–99.
Heinze J, Buschinger A, Poettinger T, Suefuji M. Multiple convergent origins of workerlessness and inbreeding in the socially parasitic ant genus Myrmoxenus. PLoS One. 2015;10:e0131023.
Buschinger A. Biological and systematic relationships of social parasitic Leptothoracini from Europe and North America. Biosystematics of social insects. Systematics Association Special Volume. 1981;19:211–22.
Bolton B. Synopsis and classification of Formicidae. American Entomological Institute. 2003;71:1–361.
Ward PS. Phylogeny, classification, and species-level taxonomy of ants (hymenoptera: Formicidae). Zootaxa. 2007;1668:549–63.
Ward PS, Brady SG, Fisher BL, Schultz TR. Phylogeny and biogeography of dolichoderine ants: effects of data partitioning and relict taxa on historical inference. Syst Biol. 2010;59:342–62.
Lucky A. Molecular phylogeny and biogeography of the spider ants, genus Leptomyrmex Mayr (hymenoptera: Formicidae). Mol Phylogenet Evol. 2011;59:281–92.
Blaimer BB. Acrobat ants go global–origin, evolution and systematics of the genus Crematogaster (hymenoptera: Formicidae). Mol Phylogenet Evol. 2012;65:421–36.
Branstetter MG. Origin and diversification of the cryptic ant genus Stenamma Westwood (hymenoptera: Formicidae), inferred from multilocus molecular data, biogeography and natural history. Syst Entomol. 2012;37:478–96.
Moreau CS, Bell CD. Testing the museum versus cradle tropical biological diversity hypothesis: phylogeny, diversification, and ancestral biogeographic range evolution of the ants. Evolution. 2013;67:2240–57.
Economo EP, Klimov P, Sarnat EM, Guénard B, Weiser MD, Lecroq B, Knowles LL. Global phylogenetic structure of the hyperdiverse ant genus Pheidole reveals the repeated evolution of macroecological patterns. Proc Biol Sci. 2015;282:20141416.
Clouse RM, Janda M, Blanchard B, Sharma P, Hoffmann BD, Andersen AN, Czekanski-Moir JE, Krushelnycky P, Rabeling C, Wilson EO, Economo EP. Molecular phylogeny of indo-Pacific carpenter ants (hymenoptera: Formicidae, Camponotus) reveals waves of dispersal and colonization from diverse source areas. Cladistics. 2015;31:424–37.
Blaimer BB, Brady SG, Schultz TR, Lloyd MW, Fisher BL, Ward PS. Phylogenomic methods outperform traditional multi-locus approaches in resolving deep evolutionary history: a case study of formicine ants. BMC Evol Biol. 2015;15:271.
Ward PS, Fisher BL. Tales of dracula ants: the evolutionary history of the ant subfamily Amblyoponinae (hymenoptera: Formicidae). Syst Entomol. 2016;41:683–93.
Boudinot BE, Probst RS, Brandao CR, Feitosa RM, Ward PS. Out of the Neotropics: newly discovered relictual species sheds light on the biogeographical history of spider ants (Leptomyrmex, Dolichoderinae, Formicidae). Syst Entomol. 2016;41(3):658–71.
Ward PS, Branstetter MG. The acacia ants revisited: convergent evolution and biogeographic context in an iconic ant/plant mutualism. Proc Biol Sci. 2017;284:20162569.
Bernard F. Faune de l'Europe et du Bassin Méditerranéen. 3. Les fourmis (Hymenoptera Formicidae) d'Europe occidentale et septentrionale. Paris: Masson; 1967.
Brikiatis L. The de Geer, Thulean and Beringia routes: key concepts for understanding early Cenozoic biogeography. J Biogeogr. 2014;41:1036–54.
Marincovich L, Gladenkov AY. Evidence for an early opening of the Bering Strait. Nature. 1999;397:149–51.
Iturralde-Vinent MA, MacPhee RD. Paleogeography of the Caribbean region: implications for Cenozoic biogeography. American museum of natural history; 1999.
Emery C. Über den Ursprung der dulotischen, parasitischen under myrmekophilen Ameisen. Biol Zent Bl. 1909;11:352–62.
Bourke AF, Franks NR. Alternative adaptations, sympatric speciation and the evolution of parasitic, inquiline ants. Biol J Linnean Soc. 1991;43:157–78.
Savolainen R, Vepsäläinen K. Sympatric speciation through intraspecific social parasitism. Proc Natl Acad Sci U S A. 2003;100:7169–74.
Vepsäläinen K, Ebsen JR, Savolainen R, Boomsma JJ. Genetic differentiation between the ant Myrmica rubra and its microgynous social parasite. Insect Soc. 2009;56:425–37.
Rabeling C, Schultz TR, Pierce NE, Bacci M. A social parasite evolved reproductive isolation from its fungus-growing ant host in sympatry. Curr Biol. 2014;24:2047–52.
Seifert B, Buschinger A, Aldawood A, Antonova V, Bharti H, Borowiec L, Dekoninck W, Dubovikoff D, Espadaler X, Flegr J, Georgiadis C, Heinze J, Neumeyer R, Ødegaard F, Oettler J, Radchenko A, Schultz R, Sharaf M, Trager J, Vesnić A, Wiezik M, Zettel H. Banning paraphylies and executing Linnaean taxonomy is discordant and reduces the evolutionary and semantic information content of biological nomenclature. Insect Soc. 2016;63:237–42.
Ward PS, Brady SG, Fisher BL, Schultz TR. 2016. Phylogenetic classifications are informative, stable, and pragmatic: the case for monophyletic taxa. Insect Soc. 2016;63:489–92.
Darwin C. The origin of species by means of natural selection, or the preservation of favored races in the struggle for life. London: John Murray; 1859.
Forel A. Les Fourmis de la Suisse, Systématique, notices anatomiques et physiologiques, architecture, distribution géographique, nouvelles expériences et observations de moeurs. Zurich: H. Georg; 1874.
Wheeler WM. The social insects: their origin and evolution. London: Paul, Trench, Trubner & Co.; 1928.
Wilson EO, Hölldobler B. The rise of the ants: a phylogenetic and ecological explanation. Proc Natl Acad Sci U S A. 2005;102:7411–4.
Lucky A, Trautwein MD, Guenard BS, Weiser MD, Dunn RR. Tracing the rise of ants-out of the ground. PLoS One. 2013;8:e84012.
Baroni UC. Phylogeny and behavioural evolution in ants, with a discussion of the role of behaviour in evolutionary processes. Ethol Ecol Evol. 1989;1:137–68.
Ward PS, Downie DA. The ant subfamily Pseudomyrmecinae (hymenoptera: Formicidae): phylogeny and evolution of big-eyed arboreal ants. Syst Entomol. 2005;30:310–35.
Weisrock DW, Smith SD, Chan LM, Biebouw K, Kappeler PM, Yoder AD. Concatenation and concordance in the reconstruction of mouse lemur phylogeny: an empirical demonstration of the effect of allele sampling in phylogenetics. Mol Biol Evol. 2012;29:1615–30.
Katoh K, Standley DM. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol Biol Evol. 2013;30:772–80.
Wu M, Chatterji S, Eisen JA. Accounting for alignment uncertainty in phylogenomics. PLoS One. 2012;7:e30288.
Fernández R, Edgecombe GD, Giribet G. Exploring phylogenetic relationships within Myriapoda and the effects of matrix composition and occupancy on phylogenomic reconstruction. Syst Biol. 2016;65:871–89.
Kück P, Struck TH. BaCoCa–A heuristic software tool for the parallel assessment of sequence biases in hundreds of gene and taxon partitions. Mol Phylogenet Evol. 2014;70:94–8.
Lanfear R, Calcott B, Ho SYW, Guindon S. PartitionFinder: combined selection of partitioning schemes and substitution models for phylogenetic analyses. Mol Biol Evol. 2012;29:1695–701.
Stamatakis A. RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics. 2006;22:2688–90.
Nguyen LT, Schmidt HA, von Haeseler A, Minh BQ. Iq-tree: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies. Mol Biol Evol. 2015;32:268–74.
Ronquist F, Teslenko M, van der Mark P. MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Syst Biol. 2012;61:539–42.
Rambaut A, Suchard MA, Xie D, Drummond AJ. Tracer v1.6. 2014. http://tree.bio.ed.ac.uk/software/tracer. Accessed 1 April 2014.
Ané C, Larget B, Baum DA, Smith SD, Rokas A. Bayesian estimation of concordance among gene trees. Mol Biol Evol. 2007;24:412–26.
Larget B, Kotha SK, Dewey CN, Ané C. BUCKy: gene tree/species tree reconciliation with the Bayesian concordance analysis. Bioinformatics. 2010;26:2910–1.
Faircloth BC, Branstetter MG, White ND, Brady SG. Target enrichment of ultraconserved elements from arthropods provides a genomic perspective on relationships among hymenoptera. Mol Ecol Res. 2015;15:489–501.
Branstetter MG, Longino JT, Ward PS, Faircloth BC. Enriching the ant tree of life: enhanced UCE bait set for genome-scale phylogenetics of ants and other hymenoptera. Methods Ecol Evol. 2017;8:768–76.
Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, Adiconis X, Fan L, Raychowdhury R, Zeng Q, Chen Z, Mauceli E, Hacohen N, Gnirke A, Rhind N, di Palma F, Birren BW, Nusbaum C, Lindblad-Toh K, Friedman N, Regev A. Full-length transcriptome assembly from RNA-Seq data without a reference genome. Nature Biotechnol. 2011;29:644–52.
Borowiec ML. AMAS: a fast tool for alignment manipulation and computing of summary statistics. PeerJ. 2016;4:e1660.
Lanfear R, Frandsen PB, Wright AM, Senfeld T, Calcott B.. PartitionFinder 2: new methods for selecting partitioned models of evolution for molecular and morphological phylogenetic analyses. Mol Biol Evol. 2016:msw260.
Frandsen PB, Calcott B, Mayer C, Lanfear R. Automatic selection of partitioning schemes for phylogenetic analyses using iterative k-means clustering of site rates. BMC Evol Biol. 2015;15:13.
Guindon S, Dufayard JF, Lefort V, Anisimova M, Hordijk W, Gascuel O. New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0. Syst Biol. 2010;59:307–21.
Borowiec ML. Convergent evolution of the army ant syndrome and congruence in big-data phylogenetics. bioRxiv. 2017. doi: 10.1101/134064.
Abascal F, Zardoya R, Telford MJ. TranslatorX: multiple alignment of nucleotide sequences guided by amino acid translations. Nucleic Acids Res. 2010;38:7–13.
Larsson A. AliView: a fast and lightweight alignment viewer and editor for large data sets. Bioinformatics. 2014;30:3276–8.
Aberer AJ, Kobert K, Stamatakis A. ExaBayes: massively parallel Bayesian tree inference for the whole-genome era. Mol Biol Evol. 2014;31:2553–6.
Xie W, Lewis PO, Fan Y, Kuo L, Chen MH. Improving marginal likelihood estimation for Bayesian phylogenetic model selection. Syst Biol. 2011;60:150–60.
Bergsten J, Nilsson AN, Ronquist F. Bayesian tests of topology hypotheses with an example from diving beetles. Syst Biol. 2013;62:660–73.
Bouckaert R, Heled J, Kühnert D, Vaughan T, Wu C-H, Xie D, Suchard MA, Rambaut A, Drummond AJ. BEAST 2: a software platform for Bayesian evolutionary analysis. PLoS Comput Biol. 2014;10:e1003537.
Matzke N. BioGeoBEARS: BioGeography with Bayesian (and likelihood) evolutionary analysis in R scripts. Berkeley, Berkeley, CA: University of California; 2013.
Iturralde-Vinent MA. A brief account of the evolution of the Caribbean seaway: Jurassic to present. In: Prothero D, Ivany L, Nesbitt E, editors. From greenhouse to icehouse: the marine Eocene–Oligocene transition: New York. NY: Colombia University Press; 2003. p. 386–96.
Beaulieu JM, O'Meara BC. Detecting hidden diversification shifts in models of trait-dependent speciation and extinction. Syst Biol. 2016;65:583–601.
Kass RE, Raftery AE. Bayes factors. J Am Stat Assoc. 1995;90:773–95.
Deyrup M, Davis L, Cover S. Exotic ants in Florida. Trans Am Entomol Soc. 2000;126:293–326.
Wittenborn D, Jeschke JM. Characteristics of exotic ants in North America. NeoBiota. 2011;10:47–64.
Morrison LW. A review of Bahamian ant (hymenoptera: Formicidae) biogeography. J Biogeogr. 1998;25:561–71.
Deyrup M, Davis L, Buckner S. Composition of the ant fauna of three Bahamian islands. In Proceedings of the 7th symposium of natural history of the Bahamas. Bahamian Field Station, San Salvador, Bahamas; 1998. p. 23–31.
Wetterer JK. Geographic distribution of Temnothorax allardycei (hymenoptera: Formicidae). Trans Am Entomol Soc. 2017;143:73–7.
Zhang Z, Ramstein G, Schuster M, Li C, Contoux C, Yan Q. Aridification of the Sahara desert caused by Tethys Sea shrinkage during the late Miocene. Nature. 2014;513:401–4.
Zachos J, Pagani M, Sloan L, Thomas E, Billups K. Trends, rhythms, and aberrations in global climate 65 ma to present. Nature. 2001;292:686–93.
McKenna MC. Cenozoic Paleogeography of North Atlantic land bridges. In: MHP B, Saxov S, Talwani M, Thiede J, editors. Structure and development of the Greenland Scotland bridge: new concepts and methods. New York: Plenum; 1983.
Tiffney BH. The Eocene North Atlantic land bridge: its importance in tertiary and modern phytography of the northern hemisphere. J Arnold Arbor. 1985;66:243–73.
McKenna MC. Holarctic landmass rearrangement, cosmic events, and Cenozoic terrestrial organisms. Ann Missouri Bot. 1983;70:459–89.
Tiffney BH. Perspectives on the origin of the floristic similarity between eastern Asia and eastern North America. J Arnold Arbor. 1985;66:73–94.
Bralower TJ, Zachos JC, Thomas E, Parrow M, Paull CK, Kelly DC, Silva IP, Sliter WV, Lohmann KC. Late Paleocene to Eocene paleoceanography of the equatorial Pacific Ocean: stable isotopes recorded at ocean drilling program site 865. Allison Guyot Paleoceanography. 1995;10:841–65.
Kennett JP, Stott LD. Abrupt deep sea warming, paleoceanographic changes and benthic extinctions at the end of the Paleocene. Nature. 1991;353:225–9.
Knox ROB. Correlation of the early Paleogene in northwest Europe: an overview. Geol Soc Lond, Spec Publ. 1996;101:1–11.
Koch PL, Zachos JC, Dettman DL. Stable isotope stratigraphy and paleoclimatology of the Paleogene Bighorn Basin (Wyoming, USA). Palaeogeogr Palaeoclimatol Palaeoecol. 1995;115:61–89.
Clyde WC, Gingerich PD. Mammalian community response to the latest Paleocene thermal maximum: an isotaphonomic study in the northern Bighorn Basin. Wyoming Geology. 1998;26:1011–4.
Wing SL, Alroy J, Hickey LJ. Plant and mammal diversity in the Paleocene to early Eocene of the Bighorn Basin. Palaeogeogr Palaeoclimatol Palaeoecol. 1995;115:117–55.
Beard KC, Dawson MR. Intercontinental dispersal of Holarctic land mammals near the Paleocene/Eocene boundary; paleogeographic, paleoclimatic and biostratigraphic implications. Bull Soc Geol Fr. 1999;170:697–706.
Jaramillo C, Ochoa D, Contreras L, Pagani M, Carvajal-Ortiz H, Pratt LM, Krishnan S, Cardona A, Romero M, Quiroz L, Rodriguez G. Effects of rapid global warming at the Paleocene-Eocene boundary on neotropical vegetation. Science. 2010;330:957–61.
Graham A. History of the vegetation: cretaceous (Maastichtian)-tertiary. In: F of NAE Committee, editor. Flora of North America north of Mexico, vol. 1. New York: Oxford University Press; 1993. p. 57–70.
Graham A. The age and diversification of terrestrial new world ecosystems through cretaceous and Cenozoic time. Am J Bot. 2011;98:336–51.
Eberle JJ, Greenwood DR. Life at the top of the greenhouse Eocene world- a review of the Eocene flora and vertebrate fauna from Canada’s high Arctic. Geol Soc Am Bull. 2012;124:3–23.
Hartenberger JL. An Asian Grande Coupure. Nature. 1998;394:321.
Meng J, McKenna MC. Faunal turnovers of Paleogene mammals from the Mongolian plateau. Nature. 1998;394:364–6.
Wing SL. Eocene and Oligocene floras and vegetation of the Rocky Mountains. Ann Missouri Bot Gard. 1987;1987:748–84.
Bininda-Emonds OR, Cardillo M, Jones KE, MacPhee RD, Beck RM, Grenyer R, Price SA, Vos RA, Gittleman JL, Purvis A. The delayed rise of present-day mammals. Nature. 2007;446:507–12.
Wolfe JA. Late cretaceous-Cenozoic history of deciduousness and the terminal cretaceous event. Paleobiology. 1987;13:215–26.
Hines HM. Historical biogeography, divergence times, and diversification patterns of bumble bees (hymenoptera: Apidae: Bombus). Syst Biol. 2008;57:58–75.
Jansen G, Savolainen R, Vepsäläinen K. Phylogeny, divergence-time estimation, biogeography and social parasite–host relationships of the Holarctic ant genus Myrmica (hymenoptera: Formicidae). Mol Phylogenet Evol. 2010;56:294–304.
Dunai TJ, GAG L, Juez-Larré JE. Oligocene–Miocene age of aridity in the Atacama Desert revealed by exEposure dating of erosion-sensitive landforms. Geology. 2005;33:321–4.
Sun J, Ye J, Wu W, Ni X, Bi S, Zhang Z, Liu W, Meng J. Late Oligocene–Miocene mid-latitude aridification and wind patterns in the Asian interior. Geology. 2010;38:515–8.
Edinger EN, Risk MJ. Oligocene-Miocene extinction and geographic restriction of Caribbean corals: roles of turbidity, temperature, and nutrients. PALAIOS. 1994;9:576–98.
Givnish TJ, Barfuss MH, Van Ee B, Riina R, Schulte K, Horres R, Gonsiska PA, Jabaily RS, Crayn DM, Smith JAC, Winter K. Phylogeny, adaptive radiation, and historical biogeography in Bromeliaceae: insights from an eight-locus plastid phylogeny. Am J Bot. 2011;98:872–95.
Skwarra E. Ökologie der lebensgemeinschaften mexikanischer ameisenpflanzen. Zoomorphology. 1934;29:306–73.
Iturralde-Vinent MA. Meso-Cenozoic Caribbean paleogeography: implications for the historical biogeography of the region. Int Geol Rev. 2006;48:791–827.
I wish to thank Phil Ward, Brendon Boudinot, and two anonymous reviewers for their helpful comments on early drafts of this article. I especially thank Jack Longino and Michael Branstetter for coordinating projects LLAMA and ADMAC, from which many of the Neotropical specimens used in this study originated. Additional specimen donations and loans by the following people were crucial for the completion of this study: RS Anderson, H Bharti, L Borowiec, ML Borowiec, BE Boudinot, MG Branstetter, SP Cover, S Csősz, D Dubovikoff, X Espadaler, B Guenard, K Ivanov, RA Johnson, JT Longino, D Lubertazzi, CS Moreau, J Reyes, S Salata, M Staab, A Tinaut, and M Yoshimura.
Funding for this research was provided by the Department of Entomology and Nematology at UC Davis, the Jastro Shields Research Award, the Ernst Mayr travel award, generous help from PS Ward, and the National Science Foundation supported project ADMAC (DEB-1354996 and DEB-1354739).
Availability of data and materials
All raw sequence data generated by Sanger sequencing are publicly available at GenBank (see Additional file 6: Table S4 for accession numbers). Raw sequence reads of UCE loci have been deposited in the NCBI Sequence Read Archive under BioProject PRJNA393044. The full concatenated, aligned, and trimmed Sanger sequencing matrix and ‘min90’ UCE matrix and accompanying tree files are available from TreeBASE: http://purl.org/phylo/treebase/phylows/study/TB2:S21879. Additional data, including alignments, matrix partitioning files, BEAST xml files, trait and biogeography data files are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.b464m.
Ethics approval and consent to participate
Consent for publication
The author declares that they have no competing interest.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended background on the taxonomic history and fossil record of Temnothorax and relatives. (PDF 888 kb)
Extended methods. (DOCX 193 kb)
Temnothorax species groups currently used in the literature, with references. (XLSX 32 kb)
List of specimens used in this study, with collection data, depository data and unique specimen identifiers. (XLSX 59 kb)
Primers used with Sanger sequencing in this study, with references. (XLSX 86 kb)
GenBank accession numbers associated with Sanger sequences generated or otherwise used in this study. (XLSX 63 kb)
Trees inferred from the full Sanger sequencing dataset and data subsetting experiments. (PDF 437 kb)
Statistics of filtered UCE datasets used in this study. (XLSX 53 kb)
Definitions and illustrations of morphological characters used in this study. (DOCX 101 kb)
Matrix of all morphological characters and indices used in this study, by taxon. (XLSX 50 kb)
List of Temnothorax species and core formicoxenines, coded for behavioral characteristics. (XLSX 49 kb)
Results of HiSSE model comparisons for the nesting habit and parasitism datasets. (XLSX 17 kb)
Sanger dataset summary statistics. (XLSX 44 kb)
The 14 partitions and substitution models identified by PartitionFinder and used in the maximum likelihood and Bayesian analyses of the concatenated Sanger sequencing dataset. (XLSX 35 kb)
Trees inferred from single-gene Sanger sequencing analyses. (PDF 216 kb)
UCE assembly summary statistics. (XLSX 46 kb)
Trees inferred from UCE dataset subsetting experiments. (PDF 460 kb)
Chronograms inferred from calibration experiments. (PDF 636 kb)
BI trees based on morphology, and morphology + molecular data. (PDF 75 kb)
Biogeographic model comparisons from BioGeoBEARS analyses. (XLSX 44 kb)
Biogeographic reconstructions at selected nodes, inferred with BioGeoBEARS using empirical data and sensitivity analyses. (XLSX 34 kb)
Biogeographic reconstructions inferred from BioGeoBEARS anaylsis of empirical data and sensitivity analyses. (PDF 376 kb)
Character dependent diversification rates and ancestral state reconstructions inferred with HiSSE models. (PDF 1361 kb)
Support values and 95% HPDs for select nodes across analyses. (XLSX 43 kb)
About this article
Cite this article
Prebus, M. Insights into the evolution, biogeography and natural history of the acorn ants, genus Temnothorax Mayr (hymenoptera: Formicidae). BMC Evol Biol 17, 250 (2017). https://doi.org/10.1186/s12862-017-1095-8